[gmx-users] How to increase the tolerance for conjugate gradient minimization

[Erik Lindahl]

Hi Arthur,

Two answers:

It's a pretty fundamental result on numerical analysis that you can only achieve a tolerance that is sqrt(EPS), where EPS is your floating-point accuracy. If you need better tolerance you might have to try double precision.

Be careful to use e.g. switched interactions and a cutoff buffer, and you might want to emply L-BFGS minimization in the endgame.

However: this assumes you really need extremely low tolerance - the only normal application where that is the case is normal modes.

Cheers,

Erik



On Jun 4, 2010, at 7:54 PM, Arthur Roberts wrote:

> Hi, all,
> 
> Is there a way to increase the tolerance for Conjugate Gradient energy minimization?
> 
> It seems that I can only get a Tolerance (Fmax) = 1e-4
> 
> emtol doesn't seem to do the trick.  I tried several values.
> 
> Much appreciated,
> Art
> 
> Dr. Arthur Roberts, Ph.D.
> University of California, San Diego
> Skaggs School of Pharmacy and Pharmaceutical Sciences
> 9500 Gilman Drive #0703
> La Jolla, CA 92093-0703
> 
> email: aroberts99163 at yahoo.com
> cell: 206-850-7468
> skype=aroberts92122
> 
> 
> 
> -- 
> gmx-users mailing list    gmx-users at gromacs.org
> http://lists.gromacs.org/mailman/listinfo/gmx-users
> Please search the archive at http://www.gromacs.org/search before posting!
> Please don't post (un)subscribe requests to the list. Use thewww interface or send it to gmx-users-request at gromacs.org.
> Can't post? Read http://www.gromacs.org/mailing_lists/users.php
> 

----------------------------------------------------------
Erik Lindahl <lindahl at cbr.su.se>
Professor, Computational Structural Biology
Center for Biomembrane Research & Swedish e-Science Research Center
Department of Biochemistry & Biophysics, Stockholm University
Tel: +468164675 Cell: +46703844534